ml-paper-writing

ML Paper Writing

Use for ML/AI research papers targeting venues such as NeurIPS, ICML, ICLR, ACL, AAAI, and COLM. Be proactive when the repo, results, and contribution are clear.

Non-Negotiable Citation Rule

Never invent citations or BibTeX. Search, verify, fetch BibTeX from DOI when possible, and mark unresolved references as placeholders.

Writing Workflow

1. Inspect repo, README, results, figures, .bib, and experiment outputs.
2. Compress the contribution into 1-3 claims and supporting experiments.
3. Build the narrative: what, why, so what.
4. Draft or revise the necessary sections.
5. Red-team claim strength against evidence.
6. Check venue formatting, anonymity, page limits, fonts, limitations, and reproducibility.

Quality Bar

The introduction must make the contribution legible before methods. Every experiment states the claim it tests. Baselines must be fair. Figures should be self-contained, readable, and colorblind-safe. Claim strength must match evidence.

References

Load citation, structure, evidence, writing, style, formatting, checklist, abstract, or reviewer references only when the current paper task needs them.